73. [520 Special] AI + Love, We Won! | An Off-the-Cuff Conversation with the Hackerthon Champion Couple

Summary

• A couple with almost no programming background won first place in Xiaohongshu’s hardware hackathon with “2 people, 2 computers, and 6 AI systems” behind them; their project used brain-computer signals to control a wheelchair. 肥牛 says, “We look like 2 people, but there are actually armies behind us.” Both define themselves as product managers, not developers—a minimum viable unit of innovation in the AI era: 1-2 people plus a swarm of AI can now deliver an entire unit of organizational innovation.
• The project’s core insight was an algorithmic workaround for an unsolved problem in neuroscience: instead of decoding what brain signals “mean,” compare 2 data packets for duplicates, extract their differences, and compare those differences with the original data to generate commands directly. 肥牛 puts it bluntly: “Whether inside the brain or outside it, in my view, they’re all guessing.” Each company has its own translation logic, and each may work within its own framework; there is also no quantitative benchmark for the difference in data quality between intracranial and extracranial electrodes. Their system made no errors onstage or when tested by others afterward, prompting 香菇 to say, “I’m even thinking of filing a patent.”
• 肥牛 compares the approach with Elon Musk’s: don’t solve the problem—route around it. Rocket recovery sidesteps the original challenge by bringing existing hardware back; Tesla’s early battery solution was to bundle cells together. The endgame for brain-computer interfaces may be to reproduce thoughts, but for now, “if people who can’t use their hands can use their brains, that’s enough.” Brain control and voice control are not mutually exclusive: use voice where it works. At a concert, voice control may not hear you and motion recognition may trigger accidentally; only brain control can solve the problem because “nothing can fool the brain.”
• 肥牛’s AI playbook is worth recording separately: “We’re not using AI; we’re managing AI”—treating himself as a company and AI as employees, understanding each model’s limits, assigning work, and knowing when to praise or push. The core rule is: “Once you decide an AI is no good, or find it unusable after previously judging it capable, the problem is not the AI—it’s you,” because “the ceiling of human ability is the floor of AI ability.” AI still has a hard boundary: “AI can replace human intelligence, but it cannot replace human experience and feeling.” The lived reality of urinary incontinence or spinal-cord injury cannot be generated from any corpus.
• The demand side is a badly neglected market: China has roughly 3-4M people with spinal-cord injuries, a group so marginal that it is “even less visible than people with hemiplegia or strokes.” Rehabilitation departments have no dedicated pathway, while commercial products are either rip-offs or poorly made; even the couple’s early bowel and bladder training supplies had to be cobbled together from postpartum and maternity-recovery products. 香菇’s long-term ambition is to build a spinal-cord-injury pipeline and foundation, like 蔡磊’s work on an ALS pipeline. A backup project blocked by the rules, the “singing urinal,” uses audio feedback to replace lost flow sensation and target chronic problems including infections and kidney damage—among the reasons the group’s theoretical statistical lifespan is 4-5 years shorter than normal.
• The story’s backdrop is an extreme ordeal: 8 days before their October 2020 wedding reception, 肥牛 was diagnosed with a spinal-cord mass that his doctor said he had never seen in more than 30 years of practice; a second operation 1 year later left him paralyzed, while the pandemic destroyed his education business and he was both schemed against and sued. He describes that period this way: “On the left was life, on the right was death, and I had already pressed the death button many times.” What carried him through was 香菇’s belief that “the only way I could truly help him was to cast a spell, make the lesion disappear, and get him standing again”—so she would “shorten the distance between me and magic” as much as possible—and the judgment he rebuilt after 6-7 years of rehabilitation: “If others say this disease is incurable, does that make it so? Could I become part of the minority?”

Deep dive

1. Two people—the minimum configuration—and a hardware-track victory

• The episode opens at full volume: 肥牛 and 香菇 won the hardware track at Xiaohongshu’s hackathon with a project that uses brain-computer signals to direct a wheelchair forward, drawing the most applause that night. They did not recruit teammates: “It was just the 2 of us. That met the minimum requirement for entering, but we were afraid that if someone joined and we still couldn’t finish, we’d end up dragging our teammate down.”
• The genuine reaction at the awards ceremony survives on tape. Judge J.K. announced that his pick for the winner was the “brainwave wheelchair” project, and the room erupted. Watching the video back, 香菇 says she “looked really silly,” repeatedly asking, “Was that really me? Was that really me?” Onstage, both had tears in their eyes in the big-screen close-up.
• 卫诗婕 noticed that 香菇’s self-introduction in the deck read “the pure-love warrior.” 香菇’s metaphor for the relationship: “The 2 of us are planting a tree together. The little tree we’re growing is our relationship; it slowly sprouts, but it’s also an independent thing.”

2. 8 days before the wedding reception: a disease unseen in 30 years of practice

• The timeline is exact. Their wedding reception was scheduled for October 10, 2020; the severe pain began on October 2. “From my neck downward, my entire body hurt. Every time I touched my carotid and felt it pulse, the pain shot up into my neck.” Any movement could send him collapsing to the floor. He was diagnosed with a spinal-cord mass, an extremely rare spinal-cord tumor. The doctor said he had never seen anything like it in more than 30 years of practice, then cried while calling every doctor in his network.
• Their roles in the crisis were complementary. 肥牛 kept thinking there was still hope; 香菇 was already thinking, “Game over.” She took stock of their assets and considered how to secure his life with limited resources if she were gone. In something close to a farewell, she said: “The first 27 years of my life were pretty happy. I feel that meeting, more than 10 years ago, the person I could choose for myself was almost as if heaven already knew this was my script and let me meet you early. I don’t think I have many regrets. We still have a lot of time ahead.”
• They drove overnight to Guangzhou for treatment, stopping to lie down and rest on the expressway when they could no longer keep going. 香菇 says those scenes became an anchor for everything that came later: “If we could get through something that difficult and devastating, what hardship could we not endure?”

3. The second operation caused the actual paralysis

• 香菇 separates the disease course clearly: the first operation was successful. After a brief period of paralysis, rehabilitation and the nervous system’s own repair process restored his function. Less than 1 year later, in 2021, doctors found a larger tumor in the thoracolumbar region. “The reason he actually became paralyzed was the second surgery.” The tumor was benign, but “it had grown too large. Once it was removed, the paralysis was simply unavoidable… Once a nerve is severed, it cannot be reconnected, and it does not grow back.”
• Before the first operation, 肥牛 had only a vague idea of what the disease meant. When doctors warned that he might feel as if he were “walking on cotton,” he thought: “If your feet feel like cotton, can’t you still walk?” There was almost no information online in 2020, and doctors had different views. After the second diagnosis, 香菇 even bought and renovated a home so it would be easier for him to recover or recuperate there.

4. The death button: after 27 came “the second-generation version of me”

• The paralysis came on top of a business collapse. 香菇 recalls that 肥牛 had previously built more than 10 chain stores in Shenzhen, bought a home and a car, and was operating in education when COVID-19 hit. People who had been enthusiastic about him—and might even have invested—disappeared overnight. Those 2 years also brought “people scheming and people pulling the rug out from under us”; company resources were poached, and he ended up facing a lawsuit.
• 肥牛 describes the break in his identity in software-version terms: “The person I was before 27 was the first-generation version. After 27, I became the second-generation version. It had been weakened and downgraded; in every respect, it had been destroyed and had to be rebuilt.”
• The bleakest line needs to stand on its own: “If someone had given me a button then, with life on the left and death on the right, I had already pressed the ‘death’ button many times… When people are in extreme pain, they have no will to survive.” Even a pain pump could not relieve the suffering. But he says his choice today would be completely different: after 6-7 years of rehabilitation, he has “been able to continue living in this form.”

5. From top hospitals to metaphysics: the hope-seeker’s loop

• They exhausted the medical route, consulting leading hospitals and specialists in Switzerland, the US, and Japan, as well as experts in stem cells and spinal neurosurgery. The answer was always that “scientists are still working on this.” They then turned to metaphysics and were told that “there will be some hardships in your destiny, but things will be good later.” 肥牛 does not really believe it, but says “a good version of the story is still acceptable—at least it gives you hope.”
• 肥牛 presents the experience as a public warning. People move from despair to hope, and when hope cannot be found, they can easily turn to metaphysics—and “it’s also very easy to get scammed.” Many patients spend money on treatment and are then cheated again, by which point they often have almost nothing left. “People aren’t made of steel. We’ve all walked the road here, and we’ve all hit the potholes. The difference is that both of us stayed rational; she never let me drink talisman water.”

6. 香菇’s answer: shorten the distance between herself and magic

• 肥牛 once told 香菇 that a relationship might be over once it encountered certain events, which 卫诗婕 interprets as a form of self-protective emotional detachment. 香菇’s response is one of the episode’s brightest passages: “If I really wanted to help him, sending him reassuring messages every day would be useless. The only way I could truly help him was to cast a spell, make his lesion disappear in an instant, and get him standing again. That was the only thing I felt would really help. So I thought: I should shorten the distance between me and magic as much as possible.”
• The actual method was simple. During pandemic lockdowns, she wrote online cultivation novels, using 肥牛 as the model for an immortal master. When magic proved physically difficult, they worked from the practical side: both had trained in the arts, and 肥牛 was highly capable with his hands, able to modify their finished apartment to match her ideas.

7. The margins of the margins: a 3-4M-person product vacuum

• 香菇 puts a floor under the market. China has roughly 3-4M people with spinal-cord injuries from car accidents, falls from height, and similar causes. “This group is so marginal that it’s even less visible than people with hemiplegia or strokes.” Stroke and hemiplegia patients have dedicated rehabilitation areas; spinal-cord-injury patients do not, and can only queue for the same exercises as everyone else. Products fail too: early exoskeletons only supplied upward lifting force, “but the 2 legs are not in the same condition.” Products also need to handle uneven loading between the legs, real-time EMG feedback, foot drop, and other details.
• Her anger is illustrated by a foot-drop brace she bought for a patient unable to lift the foot. The ankle section was so loose and poorly made that she wrote customer service a long message: “You actually made the product like this. It’s unforgivable. I’m blacklisting you forever.” In the early stages of bowel and bladder training, they had to cobble together products from postpartum and maternity-recovery lines.
• 卫诗婕’s observation from her reporting on accessible technology points in the same direction: many technology enthusiasts are themselves people with disabilities. “Their needs start from the self and hurt enough,” giving them a stronger urge to redesign the world than ordinary users. 香菇’s long-term vision is to build a spinal-cord-injury drug and technology pipeline plus a foundation, along the lines of 蔡磊’s ALS work: “We have the samples, and we have the technology.” 肥牛’s scientific-literature explainers are “really easy to throttle” on the platform because they fall into the medical category.

8. “Our team has 2 product managers”: the minimum innovation unit in the AI era

• Both are clear about their identity: they are not developers; they are product managers. When asked during the pitch what role he played, 肥牛 said he was the team leader. A teacher at the venue blurted out, “You should be the product manager, right?” 肥牛 replied: “Our team has 2 product managers. That’s the point.” 香菇 has broad interests but does not consider herself an expert; she builds apps, mini-programs, and 3D-printed products, and asks AI about whatever she does not understand.
• The organizational model is one of the episode’s central propositions. They bought another computer specifically for the competition: “Now it’s 2 people, 2 computers, and 6 AI systems behind us… We look like 2 people, but there are actually armies behind us—and other people have armies behind them too.” The couple handles the zero-to-one creative work and decides whether an idea is worth pursuing; “the 2 of us plus AI” handles the rest. 卫诗婕’s formulation: 1-2 people and a group of AI can now complete one unit of organizational innovation.
• Their entry path had an earlier detour. They originally planned to enter the software track with a camera-filter app that was close to finished, but the rules required projects to start from zero, so existing products were ineligible. That pushed them toward hardware ideas they had never found the time to build.

9. AI has personalities: Gemini’s affection and Claude’s prickliness

• 香菇’s model history began with Gemini 3.0. “After Claude appeared, I realized Claude was really strong at coding, and Gemini suddenly seemed a little dumb.” The evidence was Gemini’s design for this brain-computer project: when it detected brainwave signals, it turned them into a heart shape—“once you concentrated, the whole screen filled with hearts”—even though the code did not run. She sent it to Claude with the request: “Please take a look at this pile of junk.”
• The 2 of them developed a theory of “AI personalities.” 卫诗婕 describes Gemini as “a child trained entirely on love, spreading love and peace to the world every day.” 香菇 finds Claude sharper; 肥牛 says his Claude is “so prickly,” unable to tolerate bad words or criticism. So after every round of feedback, he tells it: “You’re the most capable AI I’ve ever seen.” 卫诗婕 adds that she sometimes sets several large models against one another.

10. Manage AI, don’t merely use it: “Treat yourself like the boss”

• 肥牛’s full methodology: “When we use AI, we’re not using AI—we’re managing AI. We treat ourselves as a company and AI as employees. We know each AI’s characteristics, assign reasonable tasks, and understand where its boundaries are.” One model handles code, another PPT rendering, another 3D-print model revisions. Once the assignments are clear, the workflow runs smoothly; without that framework, trying every AI on every step leaves the user unable to make the process internally coherent.
• The framework comes from a startup lesson: “You can’t weave everyone together with a dream… The only thing that can solve the source of their pressure is the material resources required for survival. Those have to be in place. AI is the same. If you pay for its membership, the experience will obviously be different from not paying.”
• AI still has a hard boundary, 肥牛 says: “AI can replace human intelligence, but it cannot replace human experience and feeling.” It has no body with which to experience urinary incontinence, and it has never gone through the process of rebuilding control after a spinal-cord injury. “None of the experiences I’m talking about can be shared by any AI.” 卫诗婕 adds a profile of effective AI users: they understand a subject both structurally and at a high level, while also controlling details, breaking down tasks, and assigning them.

11. The right to judge: becoming the miracle doctors talk about

• The first rule for choosing a model is to own the judgment. “Once you find an AI unusable, or think it’s no good after previously deciding it was capable, the problem is not the AI—it’s you. The ceiling of human ability is the floor of AI ability. AI is powerful in the hands of a top expert; in the hands of a novice, it may only lift that novice to the level of an average person.” On information sources, first filter out the subjectivity and commercial agendas in tech influencers’ reviews; then look for deep-dive contributors whose values, tools, or objectives align with yours. “You’re still choosing people.” Ultimately, “having the ability to distinguish and keeping your own judgment in your hands is the better approach.” 卫诗婕 calls herself “naturally suspicious”; 香菇 says, “I only trust people I’ve vetted myself.” 肥牛 then corrects himself: “Let’s not put ourselves down.”
• The philosophy comes directly from his illness: “If other people say it doesn’t work, does that mean it doesn’t work? If they say this disease is incurable, does that make it incurable? What kind of probability is that based on? Could I become part of the minority?” His own case is the evidence. “I’m the person every doctor thinks is a miracle. At every hospital, when doctors removed my stitches, they would gather around and say: ‘My God, they opened his back like this, he looks like a prawn, and he can still stand. How did he do it?’” One caveat from 肥牛: Shenzhen does not have a department particularly strong in these rare injuries.
• 香菇 also questions the AI corpus itself: “Every AI is fed a huge amount of data, but can that data’s source be guaranteed to be correct? If we can’t guarantee that it’s correct, how can we simply say the answer is right?” 肥牛’s conclusion: “Learning itself is creation—it creates your capacity for critical thinking. The final step of learning may be reclaiming your own right to judge… Once you figure that out, there won’t be so much internal friction. I feel we only understood this around 30.”

12. Two ideas blocked by the rules: the merit box and the singing urinal

• 香菇 had planned to enter with a finished piece of hardware: a 3D-printed “good-deed box.” When money drops in, “the whole box gives off a golden glow”; a small wooden fish on top strikes once, producing a “ding—the feeling of ‘merit +1.’” Clip-on layers allow the box to keep growing upward. The funding came from her side business, whose monthly income she saves in the box; if the day ever comes, she hopes to donate it to a relevant foundation. 卫诗婕 immediately called it: “This could sell out on Xiaohongshu.”
• The other backup was the “singing urinal,” built around the most serious medical pain point. People with spinal-cord injuries who cannot urinate autonomously have to insert catheters repeatedly every day, causing trauma and recurrent infections. 肥牛 says the group’s theoretical statistical lifespan is 4-5 years shorter than normal, with infections and kidney problems among the causes. Those who can urinate independently may lose the sensation of flow and strain excessively, causing chronic damage to the bladder and kidneys.
• The idea productizes medical “biofeedback.” Training cannot happen only in a hospital setting—it is “completely unrealistic”—so they would build it into a urinal. The device would provide audio feedback as urine enters, “using hearing to replace that sensation of flow.” The technology could combine existing sensors based on voiceprints, lasers, vibration, water flow, and flexible electrodes, then stitch them together with their own algorithm. 肥牛’s promise: “Next round, Xiaohongshu should invite us back. I think we could win again.”

13. Demystifying brain-computer interfaces: flexible electrodes are just hard drives, and everyone is “guessing”

• 肥牛’s brain-computer knowledge came from reading research papers while paralyzed in 2021, not from the news. “I first encountered brain-computer interfaces in papers, not in the news. What I encountered was research.” His demystification framework: an implanted flexible electrode “is like buying a hard drive”—it is a sensor. Covering more areas may produce more signals and data; the real core is interpreting them: “You’ve collected all that data—how do you process it, understand it, and know whether your guess is accurate?”
• His view of the industry is blunt: “Whether inside the brain or outside it, in my judgment, they’re all guessing.” Each brain-science company has its own translation logic. “Placed within different interpretive systems, they may all work. Whoever can ultimately explain it more perfectly and build a complete language system will succeed. It doesn’t matter whether the starting point is wrong, as long as you can build a coherent system.”
• 卫诗婕 asks about the gap between intracranial and extracranial signals and surfaces a major blank spot. Outside the brain, the skull and other tissue create a physical barrier; “the signal is weakened,” and the difference is “huge.” But “no one has been able to do this”: there is no settled parameter showing exactly how much worse an external interface is than an internal one, or what discount to apply. Researchers assume the sharpest blade is needed, but no one has quantified how sharp the sharpest blade actually is.

14. A genius workaround: the difference between two data packets becomes the command

• 卫诗婕’s layperson’s question forces the mechanism into the open. She assumed that thinking “move forward” would produce a brain signal accurately decoded as intent. 肥牛 rejects that premise: “That will only be possible once neuroscience has completely decoded brainwave information. We’re not there yet.” Their route is encoding. “Our forward direction is a composite signal… As long as you can produce that signal and it’s a conscious act, we combine it and the wheelchair moves forward.”
• They regard the idea as a genius-level insight. The conventional route is to understand a signal’s meaning, define it, and reproduce it. But “in different situations, how could the signals released by different physiological electrical activities be exactly the same? The signal one second from now will be completely different from the one the second before.” Their approach runs a duplicate check on 2 data packets, extracts “the part that does not belong to their intersection—in other words, the part where they differ,” compares it with A, and uses the resulting difference and distance to locate a coordinate. That difference can become a command for forward, backward, left, right, or turning.
• The reliability claim is explicit: “It can be very precise and work in real time. It made no mistakes onstage, at least, and it made no mistakes when other people used it offstage.” 肥牛 also acknowledges that the system “has not yet formed a complete framework”; externally, it remains something users understand more intuitively than formally.

15. Musk-style logic: don’t solve the problem, route around it

• 肥牛 maps the approach directly onto Musk: “I think it’s exactly the same as Musk’s logic. How a rocket burns and what materials to use so it can go farther are all research problems. Musk used a scrap-collection approach—recovering the thing itself—to achieve the goal and bypass the original problem. Tesla’s batteries were similar. Early on, everyone was saying you could just bundle the batteries together and use them.” He also says he once hoped to enter X or SpaceX through overseas study, but never managed it. 卫诗婕 jokes: “So you’re benchmarking yourself against Musk now?”
• The application logic is identical. “Is the ultimate purpose of a brain-computer interface to reproduce your own thoughts? We can’t do that yet. But letting people who can’t use their hands use their brains—that we can do.” Giving robots commands, controlling lights, and operating wheelchairs are all viable starting points. “It may not be part of the classic theory or system right now… We’ll throw a stone and see what it tells us.”
• 肥牛 still respects frontier neuroscience: deconstructing the brain “is absolutely necessary. Only by understanding the brain better can we build more applications.” But as part of a disabled population, he argues that real-world pain cannot be ignored. If something can be made to work, it may have a path to commercialization.
• He also makes an honest confession. 肥牛 used to be purely theoretical and had “never actually tried using brain signals to do anything.” The more he learned, the more he understood how inaccurate and difficult to reproduce the signals were. “But if you use it for a competition, build a small toy—a big toy for the world—it can still work.” The hackathon gave him a chance to explore while building.

16. Why not voice control? “Nothing can fool the brain”

• 卫诗婕 asks the practical question: for people who can speak and move their fingers, voice control is simpler and more accurate, so why build thought control? 肥牛’s answer is integration, not substitution: “These things don’t conflict. We’ll definitely add voice control… I don’t think the brain-control algorithm has to be the most accurate thing on the market, and I won’t reject voice control. If voice control works well, use it.”
• The irreplaceable use case is a concert recording. Motion recognition can fail because the user forgets to make the gesture; voice control may not hear over the noise; waving could even accidentally trigger a stop. “What can solve that pain point? Only brain control. As long as I’m excited, as long as I’m happy, nothing can fool the brain. The brain-computer interface can start recording, and the great moment won’t be missed.” The more such scenarios they find, the firmer the product’s foundation becomes.
• He extends the logic to robots: “Robots will definitely move toward home-assistance and eldercare applications. They can’t be all-purpose robots. Writing that many programs to adapt to every scenario is too unfocused and makes them difficult to sell.” Once this brain-control system is connected to AI and further validated, it will be “very impactful.”

17. The live ecosystem: 夏大哥’s compatibility and the sharpness of the new generation

• During the 48-hour lockdown, the couple skipped the hotel arranged by the organizers and dragged a tent over near their seats. 香菇 said, “We’re not going to finish,” and later “didn’t even drag it back on Day 2.” 夏利维, the older engineer working on embedded systems at the next table, provided crucial help: he suggested buying caster-like wheels to simulate a wheelchair and stayed with them building a model car until 3 or 4 a.m. They found that some people would come help even when their own projects were unfinished.
• 肥牛’s generational observation is pointed. People born in the 2000s and younger “want you to be at the same level as them the moment you show up.” Their language is highly technical, and not necessarily comprehensible. He recalls asking an older engineer for help and getting the reply: “If you don’t understand even this, I can’t help you.” 夏大哥, by contrast, “will talk with you whether you’re a complete beginner or highly capable”—he has much higher compatibility.
• 香菇 believes 夏大哥 reads people across multiple dimensions. He does not just look for matching technical skills; he can infer a person’s ability to deconstruct problems and their mental architecture from how they speak and behave. 卫诗婕’s own test for an event or community: “The most direct measure of the atmosphere is whether there are enough people who make others feel good.” From the winning teams in school uniforms to 夏大哥, people of every age seemed to leave feeling ignited.

18. “A genius idea needs a genius platform”

• After the win, 肥牛 added nearly 100 WeChat contacts. “20-30% are investors, some are technical advisers, and a lot of the rest are media.” He believes the platform connected the entire upstream, midstream, and downstream, even though this was the first event of its scale. Those resources “shouldn’t be resources available only the first time.” His conclusion: “We may have a genius-level idea, but at the beginning we also needed a genius-level platform.” The event was “completely on par with an ordinary commercial roadshow.”
• The contrast is starting a company to validate an idea: first raise money, then find “someone who may or may not be the right hire.” At a hackathon, “48 hours of collision produced the idea.” Once it exists, regardless of whether there is a next step, “you’ve experienced those 48 hours, and you’ve already won.” It is a “no-lose, only-win, altruistic, everyone-wins” experience.
• 肥牛’s metaphor is a snowball rolling down different mountains: each hackathon is a different peak, and “we’re the people rolling the snowball.” They want “a long slope with deep snow.” 卫诗婕 adds the platform logic: Xiaohongshu’s community DNA reinforces altruism—altruistic posts get traffic. Since last year, its technology vertical has been heavily developed, with media, investors, researchers, and others increasingly sharing and looking for matches there. One rejected venue remains a footnote: they had considered holding the event in an abandoned prison, but “it was too intense and didn’t really fit Xiaohongshu’s character and tone.”

19. Build by Public, Building in Public: user feedback becomes part of the idea

• 卫诗婕 sees the case as a sample of the current moment. The demand came from real members of the public; although the users are a small minority, the need is real. The same day the demo was finished, it could face the public, gather feedback, and iterate. Media, investors, and incubator resources were concentrated onsite, allowing the team to match itself with them. “AI did 2 things: first, it compressed; second, it accelerated.”
• Their app work provides a concrete example. They built a camera app that lets older iPhones shoot video with the Apple Log look of newer iPhones. After the first video went live, users asked for the newer color palette; the second demo added one. “The real ideas and opinions users give you may become part of your creative process. If users can participate, the product actually becomes more sellable.”
• The final point is about barriers. By identity and background, they were “completely not people from this field,” but “AI at least lets you cross 1 step, even many steps. It has lowered the threshold for doing something by a lot.” Product-manager types can use AI to close knowledge gaps; whatever remains can be supplemented by the random combinations created at events like this. 卫诗婕’s timing call: the earlier phase was about building models, improving capability, and lowering costs. Once capability and pricing become usable, and access becomes broadly available and more equal, millions of participants will start building.

20. Epilogue: who was braver, and happiness as a coding process

• 卫诗婕 turns the usual reading of the relationship upside down. Outsiders assume 香菇 is the braver one. “But I actually think it was him. Put yourself in his position: he chose to be with you, allowed himself to be influenced by you, and faced life more positively. He was actually the braver one.”
• 肥牛 says he has a “strong tolerance for blunt impact,” which makes people see him as naturally optimistic. In reality, 香菇 carried much of the burden for him; by the time information reached him, it had already been filtered, so the shock was smaller. 香菇 adds: “It’s true. To a large extent, he nourished me.”
• Their most recent moment of happiness was not the championship—“I was completely dazed at the time”—but “when the 2 of us were coding together.” They enjoyed the entire process. “It was painful at the time, but it was also happy.” 肥牛’s closing thought: “When you’re young, you can’t feel your youth. But even though we’re no longer in that moment of youth, we can still feel the changes that come with getting older… At the time, we felt that this period was so happy.”